Conservation DistList Archives [Date] [Subject] [Author] [SEARCH]

Subject: Image enhancement

Image enhancement

From: Leslie Courtois <conlab>
Date: Tuesday, November 23, 2004
I have two non-conservation colleagues who are trying to enhance the
image of mechanically abraded "faded" iron gall ink on parchment.
They are imaging at very high resolution (800 ppi at 48-bit) using a
BetterLight scanning back, a 120mm Schneider Apo-Symmar lens, with
full spectrum fluorescent lights.

They have tried removing the infrared filter with no improvement in
contrast between ink and parchment. That is, the IR filter used on
the camera lens to yield only the visible image.

They have done all the standard procedures of separating out the
"green" vs the "blue and green" channels in photoshop...

I have done extensive testing of imaging with filters with a
Betterlight 4x5 scanner, specifically for the purpose of revealing
very low contrast writing and invisible writing. I have several
suggestions for your colleagues. First, since they are using a light
source that emits virtually no infrared light, it is logical that
only removing the IR blocking filter will have no effect. They
should try using tungsten lights for IR imaging, which have a high
emission for IR  and their range goes up to 1200-135nm, which is
basically the same range as the scanning back when the filter is

Second, 800ppi is a very high resolution, but you can still go higher
depending on the scanner model and the size of the area being scanned.
Also, if the scan is being made using all three color channels, the
resolution is being divided between the three.

Third, separating out color channels in Photoshop after doing a full
RGB scan is not the same as true filtering. Even if the scan is very
high resolution, when you are dealing with a very low contrast
situation, an RGB scan will produce aliasing. Then when you separate
channels, they are not true separations. You have to do a high
resolution scan just in the one color channel, I suggest blue, then
select out the blue channel after scanning. This will produce a true
separation, the same as if you use a Wratten blue lens filter.

Finally, there is a Photoshop technique that has produced amazing
results for me in several situations where filters and IR were
useless and writing was totally invisible. First do a 100%
resolution scan, full RGB, tungsten lights, normal exposure and IR
blocking filter in place. Open it in Photoshop, go to
hue/saturation, and raise the saturation to about +20 and the
lightness to about +10. Then start moving the hue slider slowly up
and down the scale until you find a place where there is an increase
in contrast, even the slightest bit. You may have to adjust
saturation several times while that is being done in order to see
something. Then change the mode to CMYK, and go to levels
adjustment. Select each color individually and adjust their levels.
When you think you have achieved the most contrast possible, convert
to grayscale and adjust levels again. This works much better than
just adjusting curves in RGB mode because it allows you to better
isolate where your relevant image data is.

Leslie Courtois
Chief Conservator
Etherington Conservation Center
at the
Library of Virginia
800 East Broad Street
Richmond, VA 23219

                  Conservation DistList Instance 18:24
               Distributed: Wednesday, November 24, 2004
                       Message Id: cdl-18-24-013
Received on Tuesday, 23 November, 2004

[Search all CoOL documents]